Benchmark Almanac

EWoK

GPT-4 Turbo Apr 2024 91.2% leads since May 2024
Not in model cards

Models decide which context makes a target statement plausible. Paired examples isolate concepts about the physical and social world.

Released
May 2024
Kind
Commonsense
Level
Everyday
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Claude 3.5 Sonnet #158 on the capability index
Last model added
October 2024
Scores refreshed
November 20th, 2026
36
Entries in its paper, 25 models we track

GPT-4 Turbo leads with 91.2% among models released by October 2024 GPT-4 Turbo leads with 91.2% among models released by October 2024

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in October 2024, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-4 Turbo Apr 2024, Two-shot binary choice 91.2%
  2. 2 Claude 3.5 Sonnet Two-shot binary choice 91.1%
  3. 3 Qwen2-72B Two-shot binary choice 90.1%
  4. 4 Claude 3 Opus Two-shot binary choice 89.3%
  5. 5 GPT-4o May 2024, Two-shot binary choice 88.6%
  6. 6 Llama 3 Instruct (70B) Two-shot binary choice 87.6%
  7. 7 GPT-4o mini Two-shot binary choice 86.4%
  8. 8 Claude 3 Sonnet Two-shot binary choice 84.8%
  9. 9 Mixtral Instruct (8x22B) Two-shot binary choice 84.2%
  10. 10 Mixtral Instruct (8x7B) Two-shot binary choice 83.7%
  11. 11 Claude 3 Haiku Two-shot binary choice 82.9%
  12. 12 Claude 3.5 Haiku Two-shot binary choice 81%
  13. 13 falcon_40b_instruct Log probabilities 80.1%
  14. 14 GPT-3.5 Turbo Jan 2024, Two-shot binary choice 79.7%
  15. 15 Mixtral 8x7B Log probabilities 78.4%
  16. 16 Falcon-40B Log probabilities 78.3%
  17. 17 Mistral 7B v0.1 Log probabilities 77.5%
  18. 18 Llama 3-70B Log probabilities 77.5%
  19. 19 Mistral Instruct v0.3 (7B) Two-shot binary choice 77.2%
  20. 20 mpt_30b_chat Log probabilities 77.1%
  21. 21 MPT-30B Log probabilities 75.7%
  22. 22 mpt_7b_chat Log probabilities 75.1%
  23. 23 Llama 3-8B Log probabilities 74.6%
  24. 24 MPT-7B Log probabilities 73.3%
  25. 25 Phi-1.5 Log probabilities 72.7%
  26. 26 Falcon-7B Log probabilities 72.3%
  27. 27 Gemma 1.1 7B IT Log probabilities 72%
  28. 28 Phi-2 Log probabilities 71.8%
  29. 29 falcon_7b_instruct Log probabilities 71.7%
  30. 30 Gemma 7B Log probabilities 71.4%
  31. 31 Gemma 2B Log probabilities 67.8%
  32. 32 GPT-2 (1.5B) Log probabilities 65.5%
  33. 33 Gemma 1.1 2B IT Log probabilities 65.4%
  34. 34 word2vec Log probabilities 54.2%
  35. 35 phi_1 Log probabilities 52.2%
  36. 36 Llama 3 Instruct (8B) Two-shot binary choice 17.1%

Source: The EWoK paper

Source: The EWoK paper

#EntryScore ±SettingDateSource
1 GPT-4 Turbo Apr 2024 91.2% Two-shot binary choice Apr 2024 Official leaderboard
2 Claude 3.5 Sonnet 91.1% Two-shot binary choice Jun 2024 Official leaderboard
3 Qwen2-72B 90.1% Two-shot binary choice Jun 2024 Official leaderboard
4 Claude 3 Opus 89.3% Two-shot binary choice Feb 2024 Official leaderboard
5 GPT-4o May 2024 88.6% Two-shot binary choice May 2024 Official leaderboard
6 Llama 3 Instruct (70B) 87.6% Two-shot binary choice Official leaderboard
7 GPT-4o mini 86.4% Two-shot binary choice Jul 2024 Official leaderboard
8 Claude 3 Sonnet 84.8% Two-shot binary choice Feb 2024 Official leaderboard
9 Mixtral Instruct (8x22B) 84.2% Two-shot binary choice Official leaderboard
10 Mixtral Instruct (8x7B) 83.7% Two-shot binary choice Official leaderboard
11 Claude 3 Haiku 82.9% Two-shot binary choice Mar 2024 Official leaderboard
12 Claude 3.5 Haiku 81% Two-shot binary choice Oct 2024 Official leaderboard
13 falcon_40b_instruct 80.1% 79 to 81 Log probabilities Official leaderboard
14 GPT-3.5 Turbo Jan 2024 79.7% Two-shot binary choice Jan 2024 Official leaderboard
15 Mixtral 8x7B 78.4% ±1 Log probabilities Dec 2023 Official leaderboard
16 Falcon-40B 78.3% 77.5 to 78.9 Log probabilities Mar 2023 Official leaderboard
17 Mistral 7B v0.1 77.5% 76.8 to 78.3 Log probabilities Sep 2023 Official leaderboard
18 Llama 3-70B 77.5% 75.9 to 78.7 Log probabilities Apr 2024 Official leaderboard
19 Mistral Instruct v0.3 (7B) 77.2% Two-shot binary choice Official leaderboard
20 mpt_30b_chat 77.1% 75.8 to 77.7 Log probabilities Official leaderboard
21 MPT-30B 75.7% 74.3 to 78.6 Log probabilities Jun 2023 Official leaderboard
22 mpt_7b_chat 75.1% 73 to 76.9 Log probabilities Official leaderboard
23 Llama 3-8B 74.6% 74 to 75.6 Log probabilities Apr 2024 Official leaderboard
24 MPT-7B 73.3% 71.6 to 74.5 Log probabilities May 2023 Official leaderboard
25 Phi-1.5 72.7% 68.6 to 75.6 Log probabilities Sep 2023 Official leaderboard
26 Falcon-7B 72.3% 71.3 to 74.4 Log probabilities Apr 2023 Official leaderboard
27 Gemma 1.1 7B IT 72% 70.8 to 73.6 Log probabilities Apr 2024 Official leaderboard
28 Phi-2 71.8% 69.6 to 77.1 Log probabilities Dec 2023 Official leaderboard
29 falcon_7b_instruct 71.7% 70.1 to 73 Log probabilities Official leaderboard
30 Gemma 7B 71.4% 69.7 to 73.4 Log probabilities Feb 2024 Official leaderboard
31 Gemma 2B 67.8% 65.7 to 70.5 Log probabilities Feb 2024 Official leaderboard
32 GPT-2 (1.5B) 65.5% 64.5 to 66.2 Log probabilities Nov 2019 Official leaderboard
33 Gemma 1.1 2B IT 65.4% 64.3 to 67.3 Log probabilities Apr 2024 Official leaderboard
34 word2vec 54.2% 53.9 to 54.7 Log probabilities Official leaderboard
35 phi_1 52.2% 51.7 to 53 Log probabilities Official leaderboard
36 Llama 3 Instruct (8B) 17.1% Two-shot binary choice Official leaderboard
#ModelScore ±SettingDateSource
1 GPT-4 Turbo Apr 2024 91.2% Two-shot binary choice Apr 2024 Official leaderboard
2 Claude 3.5 Sonnet 91.1% Two-shot binary choice Jun 2024 Official leaderboard
3 Qwen2-72B 90.1% Two-shot binary choice Jun 2024 Official leaderboard
4 Claude 3 Opus 89.3% Two-shot binary choice Feb 2024 Official leaderboard
5 GPT-4o May 2024 88.6% Two-shot binary choice May 2024 Official leaderboard
6 GPT-4o mini 86.4% Two-shot binary choice Jul 2024 Official leaderboard
7 Claude 3 Sonnet 84.8% Two-shot binary choice Feb 2024 Official leaderboard
8 Claude 3 Haiku 82.9% Two-shot binary choice Mar 2024 Official leaderboard
9 Claude 3.5 Haiku 81% Two-shot binary choice Oct 2024 Official leaderboard
10 GPT-3.5 Turbo Jan 2024 79.7% Two-shot binary choice Jan 2024 Official leaderboard
11 Mixtral 8x7B 78.4% ±1 Log probabilities Dec 2023 Official leaderboard
12 Falcon-40B 78.3% 77.5 to 78.9 Log probabilities Mar 2023 Official leaderboard
13 Mistral 7B v0.1 77.5% 76.8 to 78.3 Log probabilities Sep 2023 Official leaderboard
14 Llama 3-70B 77.5% 75.9 to 78.7 Log probabilities Apr 2024 Official leaderboard
15 MPT-30B 75.7% 74.3 to 78.6 Log probabilities Jun 2023 Official leaderboard
16 Llama 3-8B 74.6% 74 to 75.6 Log probabilities Apr 2024 Official leaderboard
17 MPT-7B 73.3% 71.6 to 74.5 Log probabilities May 2023 Official leaderboard
18 Phi-1.5 72.7% 68.6 to 75.6 Log probabilities Sep 2023 Official leaderboard
19 Falcon-7B 72.3% 71.3 to 74.4 Log probabilities Apr 2023 Official leaderboard
20 Gemma 1.1 7B IT 72% 70.8 to 73.6 Log probabilities Apr 2024 Official leaderboard
21 Phi-2 71.8% 69.6 to 77.1 Log probabilities Dec 2023 Official leaderboard
22 Gemma 7B 71.4% 69.7 to 73.4 Log probabilities Feb 2024 Official leaderboard
23 Gemma 2B 67.8% 65.7 to 70.5 Log probabilities Feb 2024 Official leaderboard
24 GPT-2 (1.5B) 65.5% 64.5 to 66.2 Log probabilities Nov 2019 Official leaderboard
25 Gemma 1.1 2B IT 65.4% 64.3 to 67.3 Log probabilities Apr 2024 Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Benchmark paper Passed 85% in May 2024, by GPT-4 Turbo
GPT-4 Turbo 91.2%

Source: The EWoK paper

Claude 3 Opus on the official leaderboard passed 85% in February 2024

Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2020 2021 2022 2023 2024 2025 Near the ceiling (85%) GPT-4 Turbo GPT-3.5 Turbo Falcon-40B GPT-2 (1.5B) 0% 20% 40% 60% 80% 100% 2020 2022 2024 Near the ceiling (85%) GPT-4 Turbo GPT-3.5 Turbo

Source: Official EWoK leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Knowledge

All 332 in Knowledge
  • MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects. Intern-S2-397B 89.8%
  • MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy. Seed1.8 92.3%
  • SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy. Qwen 3.6 Max (Preview) 73.9%
  • WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1. Tencent Hy4 preview 83.9%
  • BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly. Gemma 4 31B IT 74.4%
  • HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%

Data audit

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the benchmark’s paper (36 entries).
  • Pass Models scored 25 models we track, plus 11 other entries on its leaderboard.
  • Note Newest score The newest came 2 years ago, in October 2024, for Claude 3.5 Haiku. That’s expected for a saturated test.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is Claude 3.5 Sonnet (#158).
  • Note Headroom Saturated. The best is 91.2%, past 85% since February 2024.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 25 rows for models we track and 11 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check